Looking behind the scenes: an assessment of the interdependence of Brazilian cultural industries

How important is Brazil’s cultural industries to its economy? We provide an answer to this question by evaluating the interdependence of the cultural activities in the Brazilian production structure and its evolution over the last few years (2005 – 2009). To accomplish this, we disaggregate 13 cultural economic industries in the Brazilian inputoutput table and calculate several indexes, such as, the production multiplier, linkage indexes, fields of influence and extraction analysis. Results show that the only cultural sector with high links to other sectors in the production structure is Telecommunication, edition and news agencies and that this sector provides the greatest loss in output when removed from the economy. Moreover, the sectors Jewelry, music instruments and toys, and Manufacture of telecommunication equipment have output multipliers higher than the average of the economy.


Introduction
Cultural economics is a broadly-defined term, and there has not been an agreement on its primary focus. As pointed out by Blaug (2001) and (Towse, 2003), studies of the arts was the main interest in the beginning (1960s); however, there is an increasing number of studies that deal with other aspects of cultural activities. The cultural topic has gained more attention in recent years, and we can observe supranational institutions such as UNESCO, OECD, Eurostat and the World Bank, for example, that have indicators or statistics on cultural activities. Throsby (2004) stresses the importance of an impact analysis of cultural activities in the broader sectorial sense rather than the impact of a festival or a museum, for example, which are more common in studies of cultural economics. The author clearly states that analytical methods, such as, inputoutput, social accounting matrix and computable general equilibrium are extremely important but rare because of a lack of data.
Some of the pioneer works using input-output analysis to assess cultural activities are the works of Vaughan (1984) and for the Arts (1986). The first perform multiplier analysis for tourism activities in the United Kingdom using input-output data and a survey conducted in the country. The second is a report showing the cultural activities' numbers in the 1984 input-output data in the United States. More recently, we have the studies of DiNoto and Merk (1993) analyzing the impacts of arts activities in Idaho, Gazel and Schwer (1997) assessing the impact of the Grateful Dead performance in Las Vegas, Bryan et al. (2000) evaluating the economic impacts of cultural activities in the Welsh economy, and David and Guilhoto (2012) studying the economic importance of cultural activities in Brazil in 2006.
The Brazilian cultural industries sector is a fast-growing part of the country's economy. As per data from CEMPRE 1 and PNAD 2 , IBGE reported some statistics on cultural industries for the years 2003 to 2005 and 2007 to 2010. For the first period, there was a 32% growth on gross output, 32.5% on value-added, 19% on the number of enterprises, and 32% on the government expenditure. As for the second period, there was a 31% growth on gross output, 38% on valued-added, 9% on the number of enterprises, and 64% on the government expenditure. Moreover, the number of workers grew 13% in both periods 2003 workers only), however, decreased by 11% according to PNAD (formal and informal workers).
Motivated by the increasing importance of the cultural industries in the Brazilian economy and the lack of studies using input-output analysis to assess the cultural industries, the objective of this paper is to evaluate the interdependence of the cultural activities in Brazilian production structure, therefore shedding more light on the cultural sectors feature in Brazil over time (2005 -2009). In this sense, we contribute to the literature by providing an input-output framework with the cultural industries specified in the production structure and by analyzing the importance of such industries in the Brazilian production structure. To do this we disaggregate the cultural economic sectors Brazilian input-output tables for such years, calculate the production multiplier, linkage indexes, fields of influence and extraction analysis. It is important to stress that this work differentiates itself from David and Guilhoto (2012) by adding the time effect, using the extraction analysis and mainly by how we calculate the cultural sectors in the Brazilian production structure.
The remainder of the paper is as follows: section 2 presents a brief review of the cultural studies performed for the Brazilian economy, section 3 describes the data used and its manipulation, section 4 presents the several methodologies used and their results, and in section 5 we make our conclusions.

Cultural industry in Brazil
There are several studies analyzing the cultural economic aspects of the Brazilian economy. Section 2.1 provides a brief literature review of these studies to characterize the cultural industry in Brazil. Section 2.2 presents some descriptive analysis of the cultural industry in Brazil, focusing on regional differences. Silva et al. (2007) used the RAIS 3 database to calculate the effects of economic cycles in the formal cultural labor market for the period 1995 to 2002. The authors concluded that the cultural labor market follows the same pattern as the GDP, however, it is more sensitive than sectors identified as more 4/24 "traditional." Ferreira Neto et al. (2012) assessed the wage differential among the cultural sector workers and artists, and other workers, using a pseudopanel analysis between 2002 and 2007 and found that cultural workers and artists are better paid in Brazil, per se. Machado et al. (2013b) studied the effects of public expenditure in cultural activities on creative workers' wages using data from PME 4 and FINBRA for the years 2002-2010. Golgher (2011) used the Brazilian Census to describe the formation of creative and entertainment clusters based on where skilled workers in the creative industry are located by testing Florida's hypothesis that creative and skilled workers are attracted to places with more entertainment options and diverse people. Machado et al. (2013a) analyzed the potential of creativity of Brazilian regions and municipalities based on "cultural amenities" that might influence the decision of living and working in a specific city and found three well defined clusters. The first comprises São Paulo and Rio de Janeiro (the two largest metropolises in Brazil); the second comprises state capitals and the so-called university towns; and the third cluster, having 99 municipalities, is denominated centers of cultural and ecological tourism. Ferreira Neto and Perobelli (2013) created a comparable index to evaluate the cultural activity aspects of the Minas Gerais state micro-regions and performed some exploratory spatial analysis to find possible cultural clusters in the state. The authors found that most of the micro-regions had a low Development Potential of Cultural Activities (64 out of 66 regions) and that there were some clustering patterns. Porsse et al. (2009) evaluated the economic impacts of the Natal Luz festival in Brazil's Rio Grande do Sul state using surveys and input-output techniques. The authors concluded that the festival had high impact on the overall indexes and the main reason for this is the attraction of tourists. David and Guilhoto (2012) studied the cultural sectors in the Brazilian inputoutput table for 2006, and their main findings were that the Television Industry and Journal, Magazine and Discs sectors are key-sectors in the Brazilian economy and that the employment and wage multipliers for cultural sectors are above the Brazilian average.

Cultural studies in Brazil
With respect to consumption of cultural goods, most studies use the POF 5 survey conducted by the Brazilian Institute of Geography and Statis-5/24 tics (IBGE). Diniz and Machado (2011) focus on the determinants of consumption of artistic-cultural goods in the metropolitan regions of Brazil. Earp (2009), and focus on the demand for movies in Brazil. Other than those papers, we can highlight the work of Bertini (2009) , Reis (2002Reis ( , 2006, Valiati and Florissi (2007), as well as the several reports generated by IBGE and the Ministry of Culture.

Cultural studies in Brazil
IBGE, the Brazilian Institute of Geography and Statistics, periodically releases reports on several Brazilian industries that comprise information on labor, government expenditure, and other data they annually gather. For the cultural industry, the most recent report is the Sistema de Informações e Indicadores Culturais 2007-2010. Table 1 presents summary information on other accounts included in the input-output tables. In general, the numbers are almost constants and shows that the Gross Output of the cultural industry represents around 11% of the Brazilian GDP, 12% of the Value Added, with government expenditures around 0.3% for cultural activities. Table 2 illustrates the per capita expenditure on culture and the state government expenditure on culture by Brazilian macro-regions. At the national level, the per capita expenditure and the state expenditure both increased between 2007 and 2010, the former by 77.5% and the latter by 58.5%. For the per capita expenditure, the North region has the smallest increase (22.5%) while the Center-West has the largest increase (66.8%). However, in absolute terms, the South region is where people spend less on cultural goods, R$ 22.50 as of 2010, and the Center-West is where they spend more, R$ 90.74. As for the state expenditure, the Northeastern and the Center-West regions show levels of increases similar to the national level, 73.5% and 70.4% respectively. However, the Southeast region is still the one that spends the most on culture, representing 53.3% of the national expenditure. Table 3 contains information by Brazilian macro-region about the labor force and their income. Tables 1 and 2 show that the cultural industry in Brazil is producing more (gross output and value-added) and the government is investing more (national and state expenditure), however, during 2007 and 2011 there is a decline in the number of workers in the cultural industry in almost every region; the exception is the North region. On the other hand, the average income both in total and in the cultural industries has increased during the period.

Data
To assess the interdependence of the Brazilian cultural industries, we use estimated input-output tables (Guilhoto andSesso-Filho (2005, 2010)) 6 , with 7/24 56 industries (Appendix 1) and 110 commodities for 2005 and 2009. From the 56 industries, however, the Brazilian Institute of Geography and Statistics (IBGE) considers only 2 as cultural activities: Public and Private Education.
Nonetheless, the cultural activities have a much larger participation in the production structure. To disaggregate such activities, we used information from the Annual Service Survey, Annual Industry Survey and the System of Information and Cultural Indexes 7 from IBGE. These surveys provide information to disaggregate some of the industries, enabling us to identify 13 cultural sectors.
We followed four steps to disaggregate and estimate the input-output tables with the cultural sectors 8 . First, we identified the cultural activities and their input-output corresponding sector. Secondly, we summed the gross output of cultural activities provided by the Survey and got the share of cultural activity in the corresponding I-O sector. Thirdly, from this share we disaggregated the I-O table and then used the RAS method to rebalance the Input-Output Table. Table 4 presents this share of cultural activities in their respective inputoutput industry. From the 13 industries that contain some sort of cultural activities, the ratio for cultural -non-cultural is generally similar between 2005 and 2009. However, for 2 industries -31 and 33 -this ratio changes from around 30% cultural in 2005 in both industries to 72% and 92.5% cultural in 2009, respectively. We believe that this change is due to the classification system used in Brazil. In 2005, the information provided for activity used CNAE 9 1.0 and in 2009 CNAE 2.0. Although both classifications are compatible, issues like this are likely to happen and do not bias our results, hence we need some extra care to analyze the results for these sectors.
Appendices 2 and 3 present each cultural activity from the Annual Industry Survey used and their respective input-output industry for 2005 and 2009. Appendix 4 presents the activities from the Annual Service Survey used and their respective input-output industry as well.
Thus, the initial 56-industry input-output 4 Input-output model The input-output matrix describes the intersectoral flows in an economy used to analyze the industrial interdependence in an economy Miller and Blair (2009). Let the following equation represent an input-output matrix describing the monetary flows of an economy: where, Z is a matrix that represents intermediate consumption, f is the vector of final demand, and x is the vector of gross output.
where each element of A is defined as a ij = z ij /x j , which corresponds to the proportion of input that industry j needs from industry i to produce $1 of product;x is the diagonal matrix with the elements of vector x.
Solving (1), we have: after algebraic manipulations we get: where L is the Leontief inverse matrix, defined as L = (I − A) −1 = [l i j], and its elements can be understood as the direct and indirect requirements of industry j for meeting a unit of output growth in i. Sections 4.1 to 4.4 each present one of the methodologies used to analyze the interdependence of the cultural industries in Brazil. Section 4.1 presents the production multiplier, section 4.2 the linkage indexes, section 4.3 the fields of influence and section 4.4 the extraction method. Miller and Blair (2009) define the output multiplier (O j ) for sector j as the direct and indirect quantity of output required to satisfy a unity increase in the final demand of industry j. The higher the output multiplier, the higher the impact in the overall economy due to a shock in sector j. Formally, we have:

10/24
In order to evaluate the importance of the cultural industries, we compare its aggregate with respect to other central industries as well as the relative importance of each cultural industry. In this sense, table 5 presents the output multiplier for the years 2005 and 2009 for some aggregated sectors, while Figure 1 presents the output multiplier for each cultural industry.
As we can see in Table 6, the only two sectors with a higher output multiplier than the average (1.89 and 1.83) are Extractive (1.96 and 1.92) and Industry (2.11 and 2.02). Although the cultural sector multiplier is not above the mean, it is higher than the multiplier for Trade, Services and Public Administration. David and Guilhoto (2012) find similar results and point out that this is because of several problems in the production chain of cultural activities.
We can see from Figure 1

Linkage Indexes
The key-sector analysis is one of the most traditional measures to determine if a sector, or group of sectors, is important in the production structure of a country. The idea is to find out which sectors have more linkages within an economy. The most traditional manner to do so is through the Hirschman-Rasmussen indexes (HRI). According to Miller and Blair (2009), the backward linkage (U j ) provides the dependence of sector j from all the others sectors of the economy, while the forward linkage (U i ) provides the depen-

11/24
dence of all the other sectors of the economy from sector j. Values higher than one for the indexes is evidence that the sector is above the national mean. Mathematically, where [g ij ] = G = (I − K) −1 , is the element of the Ghosh (G) matrix, such that K = [k ij ] is the technical coefficient matrix defined by k ij = z ij /x i . Figure 2 shows the results for the Hirschman-Rasmussen indexes. Sectors with values above unit are considered more important to the production structure, and if both Backward Linkage (BL) and Forward Linkage (FL) are 12/24 above unit, it is considered a key sector. The only cultural industry considered a key-sector is the Printing and Publishing (13C) in 2005. Another sector that appears relevant is the 49C -Travel Agencies and related services -with a high forward linkage in both years.
As the HRI does not take into account the size of the sector in the economy,  devised a new methodology to calculate the linkage indexes using the decomposition of the Leontief inverse matrix creating the so-called Pure Linkage Indexes (PLI). Thus, the Pure Backward Linkage (PBL) yields the pure impact on the economy of the value of the total production in sector j, i.e, the impact that is free from demand for inputs from sector j and feedback from the economy to sector j and viceversa. The Pure Forward Linkage (PFL) yields the pure impact on sector j of the total production in the rest of the economy. Another advantage of the PLI is the possibility of obtaining the total index (Pure Total Linkage -PTL) through the sum of PBL and PFL. In order to make it easier to compare PLI and HRI, we can standardize the PLI, dividing the pure index of each sector by the mean of pure index in the economy.
Mathematically 10 , the Pure Linkage indexes are: where ∆ j = (I − A jj ) −1 and ∆ r = (I − A rr ) −1 . Figure 3 presents the results for the Pure Linkage Index. Different from the H-R index, if the Pure Total Linkage Normalized (PTLN) is above unit, we define it as a key sector. For the cultural industries, we have the Telecommunication (44C) sector as a key-sector in both years. The difference in the results is due to the importance of this industry in the production structure that now is being taken into account. The industries Private Education (50C) and Public Education (54C) show themselves to be important as well, especially with respect to backward effects, i.e., they demand more from the economy to meet a final demand shock than other cultural sectors.

Fields of Influence
The fields of influence analysis is complementary to the linkage indexes as it describes the distribution in changes in the technical direct coefficients in the economic system, enabling us to determine which are the most important relations in the production structure (Guilhoto (2009)). The approximation of the fields of influence from a small change in a ij is F ( ij ) such that where ij is the small change in coefficient a ij . To evaluate which coefficients have higher fields of influence we must define S ij : where S i j is the associated value to each F ( ij ) matrix, i.e., the higher S ij , the higher the field of influence of that relation in the economy. Figures 5 and 6 present the results for the field of influence for the years 2005 and 2009, respectively. The darker the link, the more important it is to the production structure. We can see that from 2005 to 2009 there is a gain in importance in the cultural industries links to the production structure, especially for sectors Manufacture of telecommunication equipment (33C), Jewelry, music instruments and toys (39C) mainly as suppliers, and Telecommunication, edition and news agencies (44C) mainly as a consumer. Figure  4 presents the key for figures 5 and 6.

17/24
The extraction method evaluates the impact of a hypothetical isolation of one region (or industry) on the rest of the economy. Therefore, this is another measurement of the linkages in the economy, complementing the previous indexes.
Mathematically, the impact vectors are and where A is the matrix of technical coefficient, L is the Leontief inverse, f is the final demand, superscript 1 represents one sector (or region), superscript R represents the rest of the economy (sectors or regions), andx is output of the rest of the economy with the extraction of the sector (or region).
Tables 7 and 8 show the result for the extraction method for the years of 2005 and 2009, respectively. Comparing the results from both tables, it is clear that the cultural industries gain in importance in 2009 compared to 2005. This can be verified in two ways: a higher impact on itself and on the other cultural industries.

Conclusions
In this paper, we examine the importance of the cultural industries in the Brazilian economy by evaluating the interdependence of the cultural activities in Brazilian production structure and its evolution over the last few years (2005 -2009) using input-output analysis. This analysis fits well with our question as the indicators calculated map several aspects of the sectorial interdependence, such as multipliers and linkages. The novelty of this approach is twofold: the disaggregated treatment of the cultural sector and the application for the Brazilian economy.
We summarize the main findings as follows: a) the output multiplierthe cultural activities are below the average but above the important sectors, such as, Trade, Services and Public Administration; b) linkages -Printing and Publishing (13C) in 2005 is classified as a key-sector and Travel Agencies and related services (49C) present the highest forward linkage in both years; c) spread effects (fields of influence) -we verify only a slight interaction between cultural activities and non-cultural activities; however, this interaction shows a slight increase in the 2009 year; and d) impact effects -the